What is in the 28-point US plan for Ukraine? Why is Europe opposing Trump's peace plan? Is the fall of Pokrovsk inevitable? 'A corruption scandal may well end the Ukraine war' Russia says talks to end Ukraine war'serious' but rules out concessions Russia says the United States-brokered talks to end the war with Ukraine are "serious", but its officials caution that an agreement is a long way off and Moscow would offer no major concessions to Kyiv. Kremlin spokesman Dmitry Peskov said in televised comments on Wednesday that the negotiations were ongoing and "the process is serious."

Continual learning is a key feature of biological neural systems, but artificial neural networks often suffer from catastrophic forgetting. Instead of backpropagation, biologically plausible learning algorithms may enable stable continual learning. This study proposes columnar-organized spiking neural networks (SNNs) with local learning rules for continual learning and catastrophic forgetting. Using CoLaNET (Columnar Layered Network), we show that its microcolumns adapt most efficiently to new tasks when they lack shared structure with prior learning. We demonstrate how CoLaNET hyperparameters govern the trade-off between retaining old knowledge (stability) and acquiring new information (plasticity). We evaluate CoLaNET on two benchmarks: Permuted MNIST (ten sequential pixel-permuted tasks) and a two-task MNIST/EMNIST setup. Our model learns ten sequential tasks effectively, maintaining 92% accuracy on each. It shows low forgetting, with only 4% performance degradation on the first task after training on nine subsequent tasks.

This paper presents a lightweight software-based approach for running spiking neural networks (SNNs) without relying on specialized neuromorphic hardware or frameworks. Instead, we implement a specific SNN architecture (CoLaNET) in Rust and optimize it for common computing platforms. As a case study, we demonstrate our implementation, called Spiffy, on a Raspberry Pi using the MNIST dataset. Spiffy achieves 92% accuracy with low latency - just 0.9 ms per training step and 0.45 ms per inference step. The code is open-source.

In the last decade, neural networks (NNs) have driven significant progress across various fields, notably in deep reinforcement learning, highlighted by studies like [1, 2, 3]. This progress has the potential to make changes in many areas such as embedded devices, IoT and Robotics. Although modern Deep Learning models have demonstrated impressive gains in accuracy, their large sizes pose limits to their practical use in many real-world applications [4]. These applications may impose requirements in energy consumption, inference latency, inference throughput, memory footprint, real-time inference and hardware costs. Numerous studies have attempted to make neural networks more efficient.

In October last year, a Russian news site published a short video of Yevgeny Prigozhin, founder of the Wagner Group, the Russian mercenary army, sitting with four men on a rooftop terrace in the resort town of Gelendzhik, on Russia's Black Sea coast. Two are missing parts of a leg. A third lost an arm. They are identified as pardoned former convicts, returned from the front in Ukraine after joining Wagner from prison. "You were an offender, now you're a war hero," Prigozhin tells one man in the clip. It was the first video to depict the return of some of the thousands of convicts who joined Wagner in return for the promise of a pardon if they survived six months of the war. Reuters news agency used facial recognition software to examine this video and more than a dozen others and photographs of homecoming convict fighters, published between October 2022 and February 2023.